miR-17~92 Promotes Progression of ABC-DLBCL Lymphoma via Regulation of Canonical NF-kB Signaling

Background: Activated B-cell like diffuse large B-cell lymphoma (ABC-DLBCL) is an aggressive lymphoma characterized by constitutive NF-κB activation. Nevertheless, the role and mechanisms of miR-17~92 in contributing to the NF-κB activation in ABC-DLBCL are still elusive. Methods: The expression of miR-17~92 primary transcript (MIR17HG) and NF-κB target genes was determined using RNA-sequencing. The expression of miR-17~92 was performed using microarray analysis. Plasmids carrying conditional over-expression and loss-of-function of miR-17~92 were respectively constructed and dual-luciferase reporter assay was used to validate the target gene of miR-17~92. Immunoprecipitation and polyubiquitination were further used to the study of potential mechanisms. Results: Expression of MIR17HG was positively correlated with NF-κB activity, miR-17~92 activated the NF-κB signaling in ABC-DLBCL, and its over-expression promoted ABC-DLBCL cell growth, accelerated cell G1 to S phase transition and enhanced cell resistance to NF-κB inhibitor. Importantly, miR-17~92 promoted NF-κB activation through directly targeting multiple ubiquitin-editing regulators to lead to increase the K63-linked polyubiquitination and decrease the K48-linked polyubiquitination of RIP1 complex in ABC-DLBCL. We further found that miR-17~92 selectively activated IκB-α and NF-κB p65 but not NF-κB p52/p100, and high miR-17~92 expression was also associated with poorer outcome in ABC-DLBCL patients. Conclusions: Taken together, miR-17~92 selectively activate the canonical NF-κB signaling via targeting ubiquitin-editing regulators to lead to constitutively NF-κB activation and poorer outcome, which is an innovative function of miR-17~92 and previously unappreciated regulatory mechanism of NF-κB activation in ABC-DLBCL. Targeting miR-17~92 may thus provide a novel bio-therapeutic


Background
Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoid malignancy, which is classi ed into germinal center B-cell-like (GCB), activated B-cell-like (ABC) and unclassi ed DLBCL based on gene expression pro ling.They have markedly different clinical features, and ABC-DLBCL has distinctly worse prognosis compared to GCB-DLBCL [1].It has been reported that constitutive NF-κB activation plays important role in lymphomagenesis of ABC-DLBCL and somatic mutations of genes, including CD79A/B, CARD11, A20, and MYD88, which are characteristically observed in ABC-DLBCL, are directly related to the constitutive NF-κB activation [2].Nevertheless, close to 50% of ABC-DLBCL do not have these characteristic genetic mutations and mechanisms of NF-κB activation in these tumors are still elusive.NF-κB activation is critical for B cell development and lymphoid organogenesis, and is mediated by homo-or hetero-dimers of NF-κB family members, including NF-κB1 (p50/p105), NF-κB2 (p52/p100), RelA (p65), RelB and Rel (c-Rel).Signaling pathways mediating NF-κB activation are primarily classi ed into canonical and non-canonical signaling, which involve the release of different NF-κB dimers from inactive cytoplasmic complexes into the nucleus [3].Canonical NF-κB signaling relies on the phosphorylation of IκBs, particularly IκB-α, leading to nuclear translocation of various NF-κB dimers, predominantly the p50/p65, however, non-canonical NF-κB signaling leads to nuclear translocation of the p52/RelB dimer through a special mechanism which relies on the inducible procession of p100 instead of IκB-α.Until now, which signaling of them mediates the constitutive NF-κB activation still remains unclear in ABC-DLBCL.In addition, the interactions between microRNAs and NF-κB signaling have been also con rmed, suggesting that dysfunction of these interactions contributes to the development of NF-κB "driven" in ammation and tumors [4,5].MicroRNA-17 ~ 92 (miR-17 ~ 92) is an oncomiR cluster consisting of six distinct microRNAs, including miR-17, miR-18a, miR-19a, miR-19b, miR-20a and miR-92, which fall into four microRNA families (miR-17, miR-18, miR-19 and miR-92).MiR-17 ~ 92 is processed from the transcript of MIR17HG (also known as C13orf25), which is frequently ampli ed in lymphoma, leukemia and some solid tumors [6,7].MiR-17 ~ 92 is essential for B-cell development and its absence leads to inhibit B-cell development at pro-B to pre-B transition [8,9], suggesting an important link between miR-17 ~ 92 and its functions during B-cell lymphopoiesis.Studies have also further shown that miR-17 ~ 92 targets distinct genes, including PHLPP2, PTEN, Bim, CDKN1A/p21, CD22, FCGR2B and Chek2, which contribute to the diverse pathways in different B-cell lymphoma subtypes [10][11][12][13].But its exact function and mechanisms in DLBCL, especially in ABC-DLBCL in which NF-κB signaling is constitutively activated, are still unclear.
Here, we uncovered that miR-17 ~ 92 selectively activated the canonical but not non-canonical NF-κB signaling in ABC-DLBCL through directly targeting multiple ubiquitin-editing proteins, which leaded to regulate the K63-and K48-linked polyubiquitination of receptor-interacting protein 1 (RIP1) complex.Activation of the canonical NF-κB signaling further promoted tumor cell proliferation and chemoresistance as well as poorer survival.Our ndings provided the rationale for targeting miR-17 ~ 92 as innovative therapeutics in ABC-DLBCL.

Patients
The diagnosis of two cohorts of 38 non-GCB and 26 ABC-DLBCL patients, who were from the Tianjin Medical University Cancer Institute and Hospital for discovery cohort and University of Nebraska Medical Center for validation cohort, respectively, was con rmed by an expert panel of hematopathologists based on WHO criteria and gene expression pro ling.The study was approved by the ethics committee of Tianjin Medical University Cancer Institute and Hospital and University of Nebraska Medical Center.The informed consent was obtained from each participant.

Cell lines and primary normal B cells
The human ABC-DLBCL (U2932, Ly3, TMD8 and Ly10) and HEK293T cell lines were obtained and cultured in their respective medium supplemented with 10%~20% serum, and primary centroblasts (CD77 + CD38 hi ), as normal B cells, were isolated and puri ed from fresh tonsils using magnetic microbeads (Miltenyi Biotec Inc., Auburn, CA, USA) as previously described [14].Details were provided in the Supplementary material.

RNA extraction, RNA-sequencing and microarray analysis
Total RNA was respectively extracted from fresh frozen tissues of 38 non-GCB DLBCL patients using a RNeasy Mini Kit (Qiagen, Hilden, Germany) for RNA-sequencing and 26 ABC-DLBCL patients using mirVana™ miRNA isolation kit (Life Technologies, Grand Island, NY, USA) for microarray analysis of miRNA expression.And then all the total RNA were treated with RNase-free DNase I for 30 min at 37°C to remove residual DNA.Details were described in the Supplementary material.

Plasmids construction and NF-κB activity assay
The Tet-on recombinant plasmids including conditional over-expression of miR-17 ~ 92, which was named as pTIP-miR-17 ~ 92, and conditional loss-of-function of miR-17 ~ 92, which was named as pTRIPZ-miR-17 ~ 92 sponge, were respectively constructed as our previously report [10].The diagram and map of these plasmids were shown in Supplementary Fig. S1.NF-κB activity was measured using the NF-κB Cignal reporter together with pTRIPZ-miR-17 ~ 92 sponge for transient cell transfection.Further details were provided in the Supplementary material.

Establishment of ABC-DLBCL cell lines with conditional over-expression or loss-of-function of miR-17 ~ 92
Virus was packaged into HEK293T cells and the virus pseudovirus particles were collected to infect the target cells to generate the ABC-DLBCL cells with conditional over-expression or loss-of-function of miR-17 ~ 92, which were further screened with puromycin and sorted by uorescent expression using a FACSVantage (Bectom Dickinson, USA).Details were described in the Supplementary material.
Wild-type and mutant Luc-Target gene-3'UTR plasmids and luciferase reporter assays Luc-Target gene-3'UTR plasmids (wild-type, WT) were constructed and the binding site's seed sequence of miR-17 ~ 92 in Luc-Target genes-3'UTR plasmids were mutated using QuikChange II XL site-directed mutagenesis kit (Agilent Technologies, Santa Clara, USA) to generate corresponding mutant plasmids (mutant, Mut).All plasmids were sequenced to con rm their indentity.The primers were listed in Supplementary Table S1.Luciferase reporter assays were used to validate the direct target genes of miR-17 ~ 92.Details were provided in the Supplementary material.

Assays of cell growth, cycle and apoptosis, and real-time quantitative PCR
Cell counting and ow cytometry (Bectom Dickinson, CA, USA) were used to evaluate the growth, cycle and apoptosis.Three major miRNA numbers including miR-17-5p, miR-19b, miR-92a in miR-17 ~ 92 were determined using RT-qPCR, of which the gene-speci c primers were purchased from Applied Biosystems (Foster City, USA).Six NF-κB downstream transcriptional target genes, including TNIP1, TNFα, NFKB2, CD83, BIRC3 and IRF1, were also determined using RT-qPCR, of which the primers were listed in Supplementary Table S2.Further details were described in the Supplementary material.Immunoprecipitation, polyubiquitination of RIP1 complex and immunoblotting RIP1 complex was immunoprecipitated from stably transduced ABC-DLBCL cell lines and the pull-down was immunoblotted to detect the K63-and K48-linked polyubiquitinations.Proteins were detected by immunoblotting.Details were described in the Supplementary material.

Survival and statistical analysis
Statistical analysis was performed using SPSS (v26.0).Survival analyses were performed using Kaplan-Meier curves and log-rank test.Two-sided P < 0.05 was considered as statistically signi cant.
To further evaluate whether miR-17 ~ 92 activated the NF-κB signaling in cells, we examined the effect of miR-17 ~ 92 on NF-κB activity.As shown in Fig. 2B, the luciferase activity of NF-κB Cignal reporter signi cantly increased after TNF-α treatment for 18h, and the increase in NF-κB luciferase activity was signi cantly attenuated (approximately 67.32%) with co-expression of miR-17 ~ 92 sponge, which suppressed miR-17 ~ 92 function (P < 0.01), suggesting miR-17 ~ 92 loss-of-function suppressed NF-κB activity in HEK293T cells.To investigate whether miR-17 ~ 92 regulated NF-κB activity in ABC-DLBCL cells, we examined the expression of miR-17 ~ 92 in them.Supplementary Fig. S2A showed that miR-17 ~ 92 cluster, including miR-17-5p, miR-19b and miR-92a, highly expressed in four ABC-DLBCL cell lines than primary normal B cells.U2932 and Ly3 had higher miR-17 ~ 92 expression, and they were selected to construct the conditional loss-of-function cells, whereas LY10 and TMD8 were selected to construct the conditional over-expression cells.These transduced cells were further con rmed by RT-qPCR (Supplementary Fig. S2B-C) and enhanced RFP or GFP assay (Supplementary Fig. S3).We then determined the expression of six NF-κB downstream transcriptional target genes in these stably transduced ABC-DLBCL cells, and found that the expression of these genes signi cantly decreased in U2932 and Ly3 cells upon miR-17 ~ 92 loss-of-function.Conversely, the levels of them signi cantly augmented in Ly10 and TMD8 cells upon miR-17 ~ 92 over-expression (Fig. 2C-D).These ndings suggested that miR-17 ~ 92 indeed activated the NF-κB signaling in multiple types of cells, including ABC-DLBCL cells.Taken together, miR-17 ~ 92 enhanced the NF-κB signaling in primary non-GCB DLBCL tumors and ABC-DLBCL cells.

MiR-17 ~ 92 down-regulates multiple ubiquitin-editing proteins and activates the canonical NF-κB signaling in ABC-DLBCL cells
We further examined the role for miR-17 ~ 92-mediated ubiquitin-editing protein translations after targeting these genes.Figure 5A displayed that miR-17 ~ 92 loss-of-function up-regulated the levels of ubiquitin-editing proteins, including A20, CYLD and Rnf11, in U2932 and Ly3 cells; whereas miR-17 ~ 92 over-expression down-regulated these protein levels in Ly10 and TMD8 cells.Because these ubiquitinediting proteins are the NF-κB negative regulators, the NF-κB signaling will be activated after downregulating these proteins by miR-17 ~ 92.Next, we investigated that which one of canonical and noncanonical NF-κB pathways would be activated by miR-17 ~ 92 in ABC-DLBCL cells.Figure 5B revealed that miR-17 ~ 92 loss-of-function decreased the protein levels of phosphorylated-IκB-α and -p65 in U2932 and Ly3 cells, whereas the level of phosphorylated-p52/p100 was not changed.Conversely, miR-17 ~ 92 over-expression upregulated phosphorylated-IκB-α and -p65 levels in Ly10 and TMD8 cells, but not the level of phosphorylated-p52/p100.These results revealed that miR-17 ~ 92 selectively activated the canonical but not non-canonical NF-κB signaling in ABC-DLBCL.
MiR-17 ~ 92 is reported to also amplify the B-cell receptor (BCR) signaling via inhibiting the ITIM proteins [11].BCR signaling is one of the NF-κB upstream pathways.We therefore selected the Ibrutinib, which is a highly selective and irreversible Bruton tyrosine kinase (BTK) inhibitor, to block the BCR signaling and further con rm the approach for miR-17 ~ 92 activating the canonical NF-κB signaling in ABC-DLBCL.Supplementary Fig. S6 revealed that Ibrutinib blocked the BCR signaling and inhibited the expression of phosphorylated-IκB-α and -p65, and miR-17 ~ 92 loss-of-function further ampli ed the inhibition effect in U2932 cells.Conversely, miR-17 ~ 92 over-expression partially reversed the inhibition effect in TMD8 cells.
These results suggested that miR-17 ~ 92 selectively activated the canonical NF-κB signaling via targeting ubiquitin-editing proteins, including A20, CYLD and Rnf11, approach but not only BCR signaling in ABC-DLBCL.

MiR-17 ~ 92 is correlated with poorer outcome in ABC-DLBCL patients
To validate the correlations between miR-17 ~ 92 expression and overall survival, we performed the microarray analysis of miRNA expression in 26 ABC-DLBCL patients treated with R-CHOP and found that patients with high miR-17 ~ 92 expression tended to have poorer overall survival; however, there were too few patients for a statistical signi cance (Fig. 6C, P > 0.05).Moreover, among 38 non-GCB DLBCL patients studied by RNA-sequencing, 34 completed clinical follow-up (range, 0.5-89.1 months).We found that patients with high miR-17 ~ 92 primary transcript expression had signi cantly poorer overall survival (Fig. 6D, P = 0.043).

Discussion
Studies have demonstrated that characteristic "driver" genes, including CD79A/B, CARD11, A20 and MYD88, frequently mutate in approximate 50% ABC-DLBCL patients to further result in the constitutive NF-κB activation [2,[26][27][28].Nevertheless, for those ABC-DLBCL patients without characteristic genetic "driver" mutations, mechanisms of NF-κB activation are still elusive.Although a report showed that miR-17 ~ 92 could induce DLBCL through regulating several pathways including BCR signaling, which further activated the NF-κB pathway [29].However, the BCR signaling is only one of the NF-κB upstream pathways.It remains poorly understood whether miR-17 ~ 92 can activate the NF-κB signaling via other upstream approaches in DLBCL, especially in ABC-DLBCL.Our results demonstrated that miR-17 ~ 92 could activate the NF-κB signaling to lead to constitutively NF-κB activation through directly targeting multiple ubiquitin-editing proteins in ABC-DLBCL, which was different from the BCR approach.Hence, our ndings are signi cant, since it implies that miR-17 ~ 92 is an important and previously unappreciated regulatory mechanism during NF-κB activation for ABC-DLBCL.
In this study, three ubiquitin-editing proteins, including TNFAIP3, CYLD and Rnf11, were preferentially predicted and validated to be the direct target genes of miR-17 ~ 92 in the process of NF-κB activation in ABC-DLBCL.However, whether other regulators, including Tax1bp1, Itch and Traf3 which are not the ubiquitin-editing proteins, are also the direct target genes of miR-17 ~ 92 to contribute to the NF-κB activation in ABC-DLBCL is unclear.It needs to be further validated in future.Besides, the binding of TNFα to TNF receptor can induce the trimerization and then recruits several proteins, including RIPs, TRADD, TRAF2/5, cIAP1/2, NEMO and others, to form complex (herein named RIPs complex).RIPs are the important and true adaptors in NF-κB signaling by interacting with upstream signaling cassettes through well-characterized protein-binding domains, and recruiting IKKα/β through NEMO binding [30].RIPs consist of seven RIP family members (RIP1 ~ 7), but RIP1 is a key regulator of NF-κB activation [30,31].After formation of RIP1 complex, cIAP1/2 and Ubc5 catalyze the RIP1 polyubiquitination.Alternatively, TRAF2 and TRAF5 can also be polyubiquitinated due to their RING domains [32,33].In this study, RIP1 antibody was applied to pull down the entire RIP1 complex and then it was con rmed to be polyubiquitinated in this process of NF-κB activation by miR-17 ~ 92 in ABC-DLBCL.It means that any members of RIP1 complex are possible to be polyubiquitinated.It is unclear that which member of RIP1 complex is indeed polyubiquitinated.Further studies using high throughput analysis of post-translational modi cations of proteins may help to answer this key question.
In addition, our previous studies using miRNA array analysis demonstrated that miR-17 ~ 92 overexpressed in various B-cell lymphomas, including ABC-DLBCL [14].In this study, we also found that four ABC-DLBCL cell lines showed high miR-17 ~ 92 level using RT-qPCR method.Another report displayed that ampli cation of C13orf25 locus on chromosome 13 was frequently detected in GCB-DLBCL but not ABC-DLBCL [34].Actually expect for gene ampli cation, miR-17 ~ 92 over-expression is also collaboratively regulated by a number of oncogenic transcription factors and histone modi cation, including MYC, E2F, Spi-1, Fli-1 and H3K4 trimethylation [7,35].Our previous study also revealed that miR-17 ~ 92 over-expression might not correlate well with the C13orf25 gene ampli cation, in which miR-17 ~ 92 expression was the highest in DHL16 cells, but without C13orf25 ampli cation based on Chip-chip assay [14].Therefore, miR-17 ~ 92 over-expression in ABC-DLBCL may be secondary to the activation of c-MYC and/or other oncogenic signaling pathways, since miR-17 ~ 92 is one of the c-MYC downstream targets [36], which is often upregulated in ABC-DLBCL.Otherwise, miR-17 ~ 92 sponge was used as miR-17 ~ 92 competitive inhibitor to investigate the miR-17 ~ 92 loss-of-function in ABC-DLBCL cells in this study.Actually, miRNA sponges have been increasingly applied in miRNA loss-of-function studies due to its high speci city and strong inhibition of target miRNAs, which suppress miRNA targets at least as strongly as chemically modi ed antisense oligonucleotides [37,38].

Conclusion
In summary, this study demonstrated that miR-17 ~ 92 selectively activated the canonical but not noncanonical NF-κB signaling via directly targeting multiple ubiquitin-editing proteins, including TNFAIP3, CYLD and Rnf11, leading to increase the K63-linked polyubiquitination and attenuate the K48-linked polyubiquitination of RIP1 complex, and activation of the canonical NF-κB signaling further promoted cell proliferation and enhanced chemoresistance in ABC-DLBCL (Fig. 6E).Our ndings imply that targeting miR-17 ~ 92 may provide a novel bio-therapeutic strategy for ABC-DLBCL patients.The regulatory effects between miR-17~92 and the K63-and K48-linked polyubiquitination of RIP1 complex, and the correlations between miR-17~92 and overall survival in ABC-DLBCL patients.a: The kinetics of K63-linked polyubiquitination level of RIP1 complex upon miR-17~92 over-expression in TMD8 cells after inducing with 1μg/ml doxycycline for 0h, 3h, 6h and 9h.Immunoblotting with RIP1 antibody con rmed equal amount of these proteins.b: Expression of K63-and K48-linked polyubiquitination of

Figure 5 The
Figure 5